Abstract
With the expansion of AI systems, both in scale and complexity, it becomes ever more urgent to bridge the gap between advancement and environmental, social, and economic sustainability. In our desire to offer a practical tool for fostering sustainable AI development, we introduce an overall checklist in guiding developers in adopting sustainable practices throughout the AI development lifecycle: the Sustainable AI Development Card. From hardware selection to model training and deployment, this card offers practical strategies in the direction of energy consumption minimisation and carbon emissions reduction, each with social equity promotion. Interposed upon recent advances in energy-efficient algorithms, hardware optimisation, and ethical AI deployment, and grounded in a systematic literature review, the research concentrates on detailing specific strategies that make alignment of AI performance with attainable sustainability objectives possible. The Sustainable AI Development Card provides how developers can minimise the sustainable risks of AI systems with no significant reduction in accuracy or scalability.
Recommended Citation
Lee, Donghyeok; Li, Yilin; Xu, Junke; Todorova, Christina; and Dehghani, Alireza, "Towards Sustainable AI Development: Challenges, Opportunities, and the Sustainable AI Development Card" (2025). UK Academy for Information Systems Conference Proceedings 2025. 22.
https://aisel.aisnet.org/ukais2025/22